p4-fast-export.py

   1 #!/usr/bin/python
   2 #
   3 # p4-fast-export.py
   4 #
   5 # Author: Simon Hausmann <hausmann@kde.org>
   6 # License: MIT <http://www.opensource.org/licenses/mit-license.php>
   7 #
   8 # TODO:
   9 #       - support integrations (at least p4i)
  10 #       - support p4 submit (hah!)
  11 #       - emulate p4's delete behavior: if a directory becomes empty delete it. continue
  12 #         with parent dir until non-empty dir is found.
  13 #
  14 import os, string, sys, time, os.path
  15 import marshal, popen2, getopt, sha
  16 from sets import Set;
  17
  18 cacheDebug = False
  19
  20 silent = False
  21 knownBranches = Set()
  22 createdBranches = Set()
  23 committedChanges = Set()
  24 branch = "refs/heads/master"
  25 globalPrefix = previousDepotPath = os.popen("git-repo-config --get p4.depotpath").read()
  26 detectBranches = False
  27 changesFile = ""
  28 if len(globalPrefix) != 0:
  29     globalPrefix = globalPrefix[:-1]
  30
  31 try:
  32     opts, args = getopt.getopt(sys.argv[1:], "", [ "branch=", "detect-branches", "changesfile=", "silent", "known-branches=",
  33                                                    "cache-debug" ])
  34 except getopt.GetoptError:
  35     print "fixme, syntax error"
  36     sys.exit(1)
  37
  38 for o, a in opts:
  39     if o == "--branch":
  40         branch = "refs/heads/" + a
  41     elif o == "--detect-branches":
  42         detectBranches = True
  43     elif o == "--changesfile":
  44         changesFile = a
  45     elif o == "--silent":
  46         silent= True
  47     elif o == "--known-branches":
  48         for branch in open(a).readlines():
  49             knownBranches.add(branch[:-1])
  50     elif o == "--cache-debug":
  51         cacheDebug = True
  52
  53 if len(args) == 0 and len(globalPrefix) != 0:
  54     if not silent:
  55         print "[using previously specified depot path %s]" % globalPrefix
  56 elif len(args) != 1:
  57     print "usage: %s //depot/path[@revRange]" % sys.argv[0]
  58     print "\n    example:"
  59     print "    %s //depot/my/project/ -- to import the current head"
  60     print "    %s //depot/my/project/@all -- to import everything"
  61     print "    %s //depot/my/project/@1,6 -- to import only from revision 1 to 6"
  62     print ""
  63     print "    (a ... is not needed in the path p4 specification, it's added implicitly)"
  64     print ""
  65     sys.exit(1)
  66 else:
  67     if len(globalPrefix) != 0 and globalPrefix != args[0]:
  68         print "previous import used depot path %s and now %s was specified. this doesn't work!" % (globalPrefix, args[0])
  69         sys.exit(1)
  70     globalPrefix = args[0]
  71
  72 changeRange = ""
  73 revision = ""
  74 users = {}
  75 initialParent = ""
  76 lastChange = 0
  77 initialTag = ""
  78
  79 if globalPrefix.find("@") != -1:
  80     atIdx = globalPrefix.index("@")
  81     changeRange = globalPrefix[atIdx:]
  82     if changeRange == "@all":
  83         changeRange = ""
  84     elif changeRange.find(",") == -1:
  85         revision = changeRange
  86         changeRange = ""
  87     globalPrefix = globalPrefix[0:atIdx]
  88 elif globalPrefix.find("#") != -1:
  89     hashIdx = globalPrefix.index("#")
  90     revision = globalPrefix[hashIdx:]
  91     globalPrefix = globalPrefix[0:hashIdx]
  92 elif len(previousDepotPath) == 0:
  93     revision = "#head"
  94
  95 if globalPrefix.endswith("..."):
  96     globalPrefix = globalPrefix[:-3]
  97
  98 if not globalPrefix.endswith("/"):
  99     globalPrefix += "/"
 100
 101 def p4File(depotPath):
 102     cacheKey = "/tmp/p4cache/data-" + sha.new(depotPath).hexdigest()
 103
 104     data = 0
 105     try:
 106         if not cacheDebug:
 107             raise
 108         data = open(cacheKey, "rb").read()
 109     except:
 110         data = os.popen("p4 print -q \"%s\"" % depotPath, "rb").read()
 111         if cacheDebug:
 112             open(cacheKey, "wb").write(data)
 113
 114     return data
 115
 116 def p4CmdList(cmd):
 117     fullCmd = "p4 -G %s" % cmd;
 118
 119     cacheKey = sha.new(fullCmd).hexdigest()
 120     cacheKey = "/tmp/p4cache/cmd-" + cacheKey
 121
 122     cached = True
 123     pipe = 0
 124     try:
 125         if not cacheDebug:
 126             raise
 127         pipe = open(cacheKey, "rb")
 128     except:
 129         cached = False
 130         pipe = os.popen(fullCmd, "rb")
 131
 132     result = []
 133     try:
 134         while True:
 135             entry = marshal.load(pipe)
 136             result.append(entry)
 137     except EOFError:
 138         pass
 139     pipe.close()
 140
 141     if not cached and cacheDebug:
 142         pipe = open(cacheKey, "wb")
 143         for r in result:
 144             marshal.dump(r, pipe)
 145         pipe.close()
 146
 147     return result
 148
 149 def p4Cmd(cmd):
 150     list = p4CmdList(cmd)
 151     result = {}
 152     for entry in list:
 153         result.update(entry)
 154     return result;
 155
 156 def extractFilesFromCommit(commit):
 157     files = []
 158     fnum = 0
 159     while commit.has_key("depotFile%s" % fnum):
 160         path =  commit["depotFile%s" % fnum]
 161         if not path.startswith(globalPrefix):
 162 #            if not silent:
 163 #                print "\nchanged files: ignoring path %s outside of %s in change %s" % (path, globalPrefix, change)
 164             fnum = fnum + 1
 165             continue
 166
 167         file = {}
 168         file["path"] = path
 169         file["rev"] = commit["rev%s" % fnum]
 170         file["action"] = commit["action%s" % fnum]
 171         file["type"] = commit["type%s" % fnum]
 172         files.append(file)
 173         fnum = fnum + 1
 174     return files
 175
 176 def isSubPathOf(first, second):
 177     if not first.startswith(second):
 178         return False
 179     if first == second:
 180         return True
 181     return first[len(second)] == "/"
 182
 183 def branchesForCommit(files):
 184     global knownBranches
 185     branches = Set()
 186
 187     for file in files:
 188         relativePath = file["path"][len(globalPrefix):]
 189         # strip off the filename
 190         relativePath = relativePath[0:relativePath.rfind("/")]
 191
 192 #        if len(branches) == 0:
 193 #            branches.add(relativePath)
 194 #            knownBranches.add(relativePath)
 195 #            continue
 196
 197         ###### this needs more testing :)
 198         knownBranch = False
 199         for branch in branches:
 200             if relativePath == branch:
 201                 knownBranch = True
 202                 break
 203 #            if relativePath.startswith(branch):
 204             if isSubPathOf(relativePath, branch):
 205                 knownBranch = True
 206                 break
 207 #            if branch.startswith(relativePath):
 208             if isSubPathOf(branch, relativePath):
 209                 branches.remove(branch)
 210                 break
 211
 212         if knownBranch:
 213             continue
 214
 215         for branch in knownBranches:
 216             #if relativePath.startswith(branch):
 217             if isSubPathOf(relativePath, branch):
 218                 if len(branches) == 0:
 219                     relativePath = branch
 220                 else:
 221                     knownBranch = True
 222                 break
 223
 224         if knownBranch:
 225             continue
 226
 227         branches.add(relativePath)
 228         knownBranches.add(relativePath)
 229
 230     return branches
 231
 232 def findBranchParent(branchPrefix, files):
 233     for file in files:
 234         path = file["path"]
 235         if not path.startswith(branchPrefix):
 236             continue
 237         action = file["action"]
 238         if action != "integrate" and action != "branch":
 239             continue
 240         rev = file["rev"]
 241         depotPath = path + "#" + rev
 242
 243         log = p4CmdList("filelog \"%s\"" % depotPath)
 244         if len(log) != 1:
 245             print "eek! I got confused by the filelog of %s" % depotPath
 246             sys.exit(1);
 247
 248         log = log[0]
 249         if log["action0"] != action:
 250             print "eek! wrong action in filelog for %s : found %s, expected %s" % (depotPath, log["action0"], action)
 251             sys.exit(1);
 252
 253         branchAction = log["how0,0"]
 254 #        if branchAction == "branch into" or branchAction == "ignored":
 255 #            continue # ignore for branching
 256
 257         if not branchAction.endswith(" from"):
 258             continue # ignore for branching
 259 #            print "eek! file %s was not branched from but instead: %s" % (depotPath, branchAction)
 260 #            sys.exit(1);
 261
 262         source = log["file0,0"]
 263         if source.startswith(branchPrefix):
 264             continue
 265
 266         lastSourceRev = log["erev0,0"]
 267
 268         sourceLog = p4CmdList("filelog -m 1 \"%s%s\"" % (source, lastSourceRev))
 269         if len(sourceLog) != 1:
 270             print "eek! I got confused by the source filelog of %s%s" % (source, lastSourceRev)
 271             sys.exit(1);
 272         sourceLog = sourceLog[0]
 273
 274         relPath = source[len(globalPrefix):]
 275         # strip off the filename
 276         relPath = relPath[0:relPath.rfind("/")]
 277
 278         for branch in knownBranches:
 279             if isSubPathOf(relPath, branch):
 280 #                print "determined parent branch branch %s due to change in file %s" % (branch, source)
 281                 return branch
 282 #            else:
 283 #                print "%s is not a subpath of branch %s" % (relPath, branch)
 284
 285     return ""
 286
 287 def commit(details, files, branch, branchPrefix, parent, merged = ""):
 288     global users
 289     global lastChange
 290     global committedChanges
 291
 292     epoch = details["time"]
 293     author = details["user"]
 294
 295     gitStream.write("commit %s\n" % branch)
 296 #    gitStream.write("mark :%s\n" % details["change"])
 297     committedChanges.add(int(details["change"]))
 298     committer = ""
 299     if author in users:
 300         committer = "%s %s %s" % (users[author], epoch, tz)
 301     else:
 302         committer = "%s <a@b> %s %s" % (author, epoch, tz)
 303
 304     gitStream.write("committer %s\n" % committer)
 305
 306     gitStream.write("data <<EOT\n")
 307     gitStream.write(details["desc"])
 308     gitStream.write("\n[ imported from %s; change %s ]\n" % (branchPrefix, details["change"]))
 309     gitStream.write("EOT\n\n")
 310
 311     if len(parent) > 0:
 312         gitStream.write("from %s\n" % parent)
 313
 314     if len(merged) > 0:
 315         gitStream.write("merge %s\n" % merged)
 316
 317     for file in files:
 318         path = file["path"]
 319         if not path.startswith(branchPrefix):
 320 #            if not silent:
 321 #                print "\nchanged files: ignoring path %s outside of branch prefix %s in change %s" % (path, branchPrefix, details["change"])
 322             continue
 323         rev = file["rev"]
 324         depotPath = path + "#" + rev
 325         relPath = path[len(branchPrefix):]
 326         action = file["action"]
 327
 328         if action == "delete":
 329             gitStream.write("D %s\n" % relPath)
 330         else:
 331             mode = 644
 332             if file["type"].startswith("x"):
 333                 mode = 755
 334
 335             data = p4File(depotPath)
 336
 337             gitStream.write("M %s inline %s\n" % (mode, relPath))
 338             gitStream.write("data %s\n" % len(data))
 339             gitStream.write(data)
 340             gitStream.write("\n")
 341
 342     gitStream.write("\n")
 343
 344     lastChange = int(details["change"])
 345
 346 def extractFilesInCommitToBranch(files, branchPrefix):
 347     newFiles = []
 348
 349     for file in files:
 350         path = file["path"]
 351         if path.startswith(branchPrefix):
 352             newFiles.append(file)
 353
 354     return newFiles
 355
 356 def findBranchSourceHeuristic(files, branch, branchPrefix):
 357     for file in files:
 358         action = file["action"]
 359         if action != "integrate" and action != "branch":
 360             continue
 361         path = file["path"]
 362         rev = file["rev"]
 363         depotPath = path + "#" + rev
 364
 365         log = p4CmdList("filelog \"%s\"" % depotPath)
 366         if len(log) != 1:
 367             print "eek! I got confused by the filelog of %s" % depotPath
 368             sys.exit(1);
 369
 370         log = log[0]
 371         if log["action0"] != action:
 372             print "eek! wrong action in filelog for %s : found %s, expected %s" % (depotPath, log["action0"], action)
 373             sys.exit(1);
 374
 375         branchAction = log["how0,0"]
 376
 377         if not branchAction.endswith(" from"):
 378             continue # ignore for branching
 379 #            print "eek! file %s was not branched from but instead: %s" % (depotPath, branchAction)
 380 #            sys.exit(1);
 381
 382         source = log["file0,0"]
 383         if source.startswith(branchPrefix):
 384             continue
 385
 386         lastSourceRev = log["erev0,0"]
 387
 388         sourceLog = p4CmdList("filelog -m 1 \"%s%s\"" % (source, lastSourceRev))
 389         if len(sourceLog) != 1:
 390             print "eek! I got confused by the source filelog of %s%s" % (source, lastSourceRev)
 391             sys.exit(1);
 392         sourceLog = sourceLog[0]
 393
 394         relPath = source[len(globalPrefix):]
 395         # strip off the filename
 396         relPath = relPath[0:relPath.rfind("/")]
 397
 398         for candidate in knownBranches:
 399             if isSubPathOf(relPath, candidate) and candidate != branch:
 400                 return candidate
 401
 402     return ""
 403
 404 def changeIsBranchMerge(sourceBranch, destinationBranch, change):
 405     sourceFiles = {}
 406     for file in p4CmdList("files %s...@%s" % (globalPrefix + sourceBranch + "/", change)):
 407         if file["action"] == "delete":
 408             continue
 409         sourceFiles[file["depotFile"]] = file
 410
 411     destinationFiles = {}
 412     for file in p4CmdList("files %s...@%s" % (globalPrefix + destinationBranch + "/", change)):
 413         destinationFiles[file["depotFile"]] = file
 414
 415     for fileName in sourceFiles.keys():
 416         integrations = []
 417         deleted = False
 418         integrationCount = 0
 419         for integration in p4CmdList("integrated \"%s\"" % fileName):
 420             toFile = integration["fromFile"] # yes, it's true, it's fromFile
 421             if not toFile in destinationFiles:
 422                 continue
 423             destFile = destinationFiles[toFile]
 424             if destFile["action"] == "delete":
 425 #                print "file %s has been deleted in %s" % (fileName, toFile)
 426                 deleted = True
 427                 break
 428             integrationCount += 1
 429             if integration["how"] == "branch from":
 430                 continue
 431
 432             if int(integration["change"]) == change:
 433                 integrations.append(integration)
 434                 continue
 435             if int(integration["change"]) > change:
 436                 continue
 437
 438             destRev = int(destFile["rev"])
 439
 440             startRev = integration["startFromRev"][1:]
 441             if startRev == "none":
 442                 startRev = 0
 443             else:
 444                 startRev = int(startRev)
 445
 446             endRev = integration["endFromRev"][1:]
 447             if endRev == "none":
 448                 endRev = 0
 449             else:
 450                 endRev = int(endRev)
 451
 452             initialBranch = (destRev == 1 and integration["how"] != "branch into")
 453             inRange = (destRev >= startRev and destRev <= endRev)
 454             newer = (destRev > startRev and destRev > endRev)
 455
 456             if initialBranch or inRange or newer:
 457                 integrations.append(integration)
 458
 459         if deleted:
 460             continue
 461
 462         if len(integrations) == 0 and integrationCount > 1:
 463             print "file %s was not integrated from %s into %s" % (fileName, sourceBranch, destinationBranch)
 464             return False
 465
 466     return True
 467
 468 def getUserMap():
 469     users = {}
 470
 471     for output in p4CmdList("users"):
 472         if not output.has_key("User"):
 473             continue
 474         users[output["User"]] = output["FullName"] + " <" + output["Email"] + ">"
 475     return users
 476
 477 users = getUserMap()
 478
 479 if len(changeRange) == 0:
 480     try:
 481         sout, sin, serr = popen2.popen3("git-name-rev --tags `git-rev-parse %s`" % branch)
 482         output = sout.read()
 483         if output.endswith("\n"):
 484             output = output[:-1]
 485         tagIdx = output.index(" tags/p4/")
 486         caretIdx = output.find("^")
 487         endPos = len(output)
 488         if caretIdx != -1:
 489             endPos = caretIdx
 490         rev = int(output[tagIdx + 9 : endPos]) + 1
 491         changeRange = "@%s,#head" % rev
 492         initialParent = os.popen("git-rev-parse %s" % branch).read()[:-1]
 493         initialTag = "p4/%s" % (int(rev) - 1)
 494     except:
 495         pass
 496
 497 tz = - time.timezone / 36
 498 tzsign = ("%s" % tz)[0]
 499 if tzsign != '+' and tzsign != '-':
 500     tz = "+" + ("%s" % tz)
 501
 502 gitOutput, gitStream, gitError = popen2.popen3("git-fast-import")
 503
 504 if len(revision) > 0:
 505     print "Doing initial import of %s from revision %s" % (globalPrefix, revision)
 506
 507     details = { "user" : "git perforce import user", "time" : int(time.time()) }
 508     details["desc"] = "Initial import of %s from the state at revision %s" % (globalPrefix, revision)
 509     details["change"] = revision
 510     newestRevision = 0
 511
 512     fileCnt = 0
 513     for info in p4CmdList("files %s...%s" % (globalPrefix, revision)):
 514         change = int(info["change"])
 515         if change > newestRevision:
 516             newestRevision = change
 517
 518         if info["action"] == "delete":
 519             continue
 520
 521         for prop in [ "depotFile", "rev", "action", "type" ]:
 522             details["%s%s" % (prop, fileCnt)] = info[prop]
 523
 524         fileCnt = fileCnt + 1
 525
 526     details["change"] = newestRevision
 527
 528     try:
 529         commit(details, extractFilesFromCommit(details), branch, globalPrefix)
 530     except:
 531         print gitError.read()
 532
 533 else:
 534     changes = []
 535
 536     if len(changesFile) > 0:
 537         output = open(changesFile).readlines()
 538         changeSet = Set()
 539         for line in output:
 540             changeSet.add(int(line))
 541
 542         for change in changeSet:
 543             changes.append(change)
 544
 545         changes.sort()
 546     else:
 547         output = os.popen("p4 changes %s...%s" % (globalPrefix, changeRange)).readlines()
 548
 549         for line in output:
 550             changeNum = line.split(" ")[1]
 551             changes.append(changeNum)
 552
 553         changes.reverse()
 554
 555     if len(changes) == 0:
 556         if not silent:
 557             print "no changes to import!"
 558         sys.exit(1)
 559
 560     cnt = 1
 561     for change in changes:
 562         description = p4Cmd("describe %s" % change)
 563
 564         if not silent:
 565             sys.stdout.write("\rimporting revision %s (%s%%)" % (change, cnt * 100 / len(changes)))
 566             sys.stdout.flush()
 567         cnt = cnt + 1
 568
 569         try:
 570             files = extractFilesFromCommit(description)
 571             if detectBranches:
 572                 for branch in branchesForCommit(files):
 573                     knownBranches.add(branch)
 574                     branchPrefix = globalPrefix + branch + "/"
 575
 576                     filesForCommit = extractFilesInCommitToBranch(files, branchPrefix)
 577
 578                     merged = ""
 579                     parent = ""
 580                     ########### remove cnt!!!
 581                     if branch not in createdBranches and cnt > 2:
 582                         createdBranches.add(branch)
 583                         parent = findBranchParent(branchPrefix, files)
 584                         if parent == branch:
 585                             parent = ""
 586     #                    elif len(parent) > 0:
 587     #                        print "%s branched off of %s" % (branch, parent)
 588
 589                     if len(parent) == 0:
 590                         merged = findBranchSourceHeuristic(filesForCommit, branch, branchPrefix)
 591                         if len(merged) > 0:
 592                             print "change %s could be a merge from %s into %s" % (description["change"], merged, branch)
 593                             if not changeIsBranchMerge(merged, branch, int(description["change"])):
 594                                 merged = ""
 595
 596                     branch = "refs/heads/" + branch
 597                     if len(parent) > 0:
 598                         parent = "refs/heads/" + parent
 599                     if len(merged) > 0:
 600                         merged = "refs/heads/" + merged
 601                     commit(description, files, branch, branchPrefix, parent, merged)
 602             else:
 603                 commit(description, filesForCommit, branch, globalPrefix, initialParent)
 604                 initialParent = ""
 605         except IOError:
 606             print gitError.read()
 607             sys.exit(1)
 608
 609 if not silent:
 610     print ""
 611
 612 gitStream.write("reset refs/tags/p4/%s\n" % lastChange)
 613 gitStream.write("from %s\n\n" % branch);
 614
 615
 616 gitStream.close()
 617 gitOutput.close()
 618 gitError.close()
 619
 620 os.popen("git-repo-config p4.depotpath %s" % globalPrefix).read()
 621 if len(initialTag) > 0:
 622     os.popen("git tag -d %s" % initialTag).read()
 623
 624 sys.exit(0)